Identification of Dominant Propagation Mechanisms ... - IEEE Xplore

2 downloads 0 Views 926KB Size Report
office building have been investigated numerically using the FDTD. Penetration through and diffraction around the corner can be neglected as reflection is seen.
Identification of Dominant Propagation Mechanisms around Corners in a Single-Floor Office Building Eva C. K. Lai*, Michael J. Neve, and Allan G. Williamson Department of Electrical and Computer Engineering, The University of Auckland Private Bag 92019, Auckland, New Zealand E-mail: [email protected], {mj.neve, ag.williamson}@auckland.ac.nz

Abstract Radio wave propagation mechanisms around concrete corners in a single-floor office building have been investigated numerically using the FDTD. Penetration through and diffraction around the corner can be neglected as reflection is seen to dominate. Simulation results show that a precise knowledge of wall permittivity may not be required, as the dominant reflected component from both concrete and drywall models are of comparable size to those obtained using a PEC wall model. These results also suggest that drywall and concrete wall could be modelled as PEC with a prediction error about 5dB and 3dB, respectively. I.

Introduction

Indoor wireless communication systems have become very popular due to their untethered nature, low cost, and broadband capability. Effective system deployment requires the provision of adequate coverage while minimising cochannel interference. Simultaneously achieving these two goals is greatly dependent on the decisions made at the time of deployment. As a result, accurate yet efficient propagation models are needed for system planning. Propagation models can be broadly divided into two types, namely empirical and deterministic. Empirical models are derived from measurements; they are accurate in similar environments where the original measurements were made, but may be unreliable when extrapolated for use in other. Deterministic models are based on analytical or numerical methods; they are accurate but frequently require large computational effort. As a result, neither model is ideally suited for use as a planning tool. A different approach that has a mechanistic basis is therefore proposed herein [1]. Our aim is to determine the local mean of the signal strength by incorporating only the dominant propagation mechanisms. These estimates of the local mean can then be used to determine outage probabilities. By eliminating the unnecessary computation of less significant components, such an approximation can be expected to be computationally efficient and accurate. This paper investigates the dominant propagation mechanisms around a corner in a single-floor office building as shown in Fig. 1. A concrete service core contains two elevators and a stairwell which is known to cause significant shadowing [2]. Section II describes the modelling strategy and the environment studied. Section III presents the simulation results, followed by conclusions in Section IV.

978-1-4244-2042-1/08/$25.00 ©2008 IEEE

II.

Modelling Strategy

The environment under investigation is a twelve-floor reinforced-concrete office block with horizontal dimensions 18.5m × 18.5m, a typical floor plan of which is shown in Fig. 1(a). The central concrete service core contains two elevators, a stairwell, and associated services. This core is surrounded by a 1.5m wide corridor and offices. A 2-D Finite-Difference Time-Domain (FDTD) method [3] has been used to study the radio wave propagation around one of the corners of this core as shown in Fig. 1(b). The FDTD simulations were conducted at 1.0GHz with a uniform transverse-magnetic (TM) polarized square grid lattice. The grid resolution λo/20 was used to minimize numerical errors, where λo is the wavelength at 1.0GHz in the densest medium i.e. concrete. Initially, only the concrete service core and the corridor with total area 9.8m × 9.8m were considered. The lattice was terminated by a uniaxial perfectly matched layer (UPML) [3,4]. The concrete service core has a wall of thickness 30cm with relative permittivity, εr=6 and conductivity, σ=50mS/m. The concrete is assumed to contain vertically-directed reinforcing steel bars modelled as perfect electric conductors (PEC) with σ=1×107S/m. Reinforcing bars are 2cm in diameter and are embedded inside the concrete with a spacing of 15cm. The elevator doors are both modelled as PEC and are assumed to be closed. The resulting field strength is averaged in a square region of 2λo × 2λo to eliminate multipath fading. (b)

(a) 18.5m

9.8m

18.5m

9.8m

Transmitting antenna Receiving antenna trajectory

Elevator door Reinforced concrete

Drywall UPML

Fig. 1. (a) Floor plan and (b) FDTD model of the environment

III.

FDTD Simulations

The first part of this investigation focused on the identification of dominant mechanisms by which radio wave propagates around the corner, and the second part considered the influence of wall permittivity on the propagation process and the path loss along the corridor.

A. Identification of dominant propagation mechanisms FDTD simulations were performed without and with the presence of the outer wall (which was assumed to be lossless drywall/plasterboard with εr=4, and σ=0S/m). Fig. 2(a) shows the path gain without the outer wall, where penetration and diffraction are the propagation mechanisms along the corridor. Fig. 2(b) shows the path gain with the outer wall, where an additional mechanism— reflection—is now present. Fig. 2(c) shows the difference between these two cases. From Fig. 2(c), it is noteworthy that the received power along the corridor is substantially increased with the presence of the outer wall, even in the vicinity of corner. It is now evident that reflection resulting from the outer wall dominates both the direct propagation through and diffraction around the corner. A modulated Gaussian pulse has also been used to estimate the impulse response, and these results suggest that 1st and 2nd order reflections have significant contributions in the area near the corner; whereas 4th and 6th order reflections are the dominant mechanisms at the mid and far section of the corridor, respectively. These findings are also consistent with results obtained by a ray-based implementation. (b)

(a)

(c) -20

10

-30

9

-40

8

-50

7

6

-60

6

5

-70

-20 9

8

-40

7

7

-50

40

-60

5

-70

4

-80

4

-80

3

-90

3

-90

3

2

-100

2

-100

2

-110

1

-110

1 6

7 8 distance, m

9

1

-120

6

7 8 distance, m

9

-120

distance, m

6

distance, m

distance, m

9

-30

8

50

30

20

5 4

10

0

6

7 8 9 distance, m

10

-10

Fig. 2. Contour plots showing the path gain in decibel (a) without outer wall, (b) with outer wall, and (c) their difference i.e. {with – without}.

B. Influence of wall permittivity The material properties of the wall are challenging to characterize accurately. However, to shed light on the influence of varying the material property of the outer wall, several cases have been considered namely (a) without the outer wall (so there is no reflection resulting from it); (b) with a PEC wall (the impinging waves are totally reflected); (c) with wall permittivity close to plastic εr=2; (d) with wall permittivity close to drywall/plasterboard εr=4; and (e) with wall permittivity close to concrete εr=6. Fig. 3 shows the effect of wall permittivity changes on the path gain along the corridor. Without the presence of the outer wall (εr=1), the received power along the corridor is dominated by direct propagation through and diffraction around the corner. In this case, the received power along the sampling trajectory drops significantly (40dB) with distance. However, in the presence of the drywall (εr=4), an additional 15dB and 30dB of

received power can be observed at 3m and 9m, respectively. It is clear that reflection is the dominant mechanism for all materials investigated in the corridor area. It is also noticeable that the path gain difference between the PEC to the concrete and drywall cases is about 3dB and 5dB, respectively. One explanation of this phenomenon is that the received power along the corridor is dominated by reflection, and the loss which is introduced by each bounce is insignificant. As a result, a considerable fraction of the impinging signals would get reflected then propagates down the corridor. -15 -20 -25

pathgain, dB

-30 -35 -40 -45

εr =1

-50

εr =2

-55

εr =4

-60

εr =6

-65 -70

PEC 0

1

2

3

4 5 6 distance, m

7

8

9

10

Fig. 3. Path gain at the receiving trajectory for various wall permittivities

IV.

Conclusions

Reflection is seen to be the dominant propagation mechanism around the corner in the environment investigated. Results suggest that the sum of reflected components could increase the received power up to 30dB when the outer wall is modelled as drywall. A maximum of 3dB and 5dB prediction error can be achieved by simply modelling the walls along the corridor as PEC instead of concrete and drywall, respectively. This is because reflection dominating the direct propagation and diffraction around the corner as well as the reflection loss is seen to be insignificant. As a result, a substantial portion of the impinging waves are being reflected off the wall and then propagate down the corridor. These findings are important toward the understanding of propagation around corners, and developing a simple yet efficient propagation model for use in system planning. References: [1] E. C. K. Lai, M. J. Neve, and A. G. Williamson, “Propagation modelling in the presence of complex structures for indoor wireless communication systems”, USNC/URSI Nat. Rad. Sci. Mtg., Jul. 2007. [2] K. S. Butterworth, K. W. Sowerby, and A. G. Williamson, “Base station placement for inbuilding mobile communication systems to yield high capacity and efficiency”, IEEE Trans. on Communications, vol. 48, No. 4, pp.658-669, Apr. 2000. [3] A. Taflove and S. C. Hagness, Computational Electrodynamics: The Finite-Difference TimeDomain Method 2ed. Boston, MA: Artech House, 2000. pp. 80-89, pp. 305-313. [4] S. D. Gedney, “An anisotropic perfectly matched layer-absorbing medium for the truncation of FDTD lattices”, IEEE Trans. on Antennas Propagat., vol. 44, No. 12, pp. 1630-1639, Dec. 1996.